An end-to-end automatic speech recognition (ASR) system includes: a first encoder configured for close-talk input captured by a close-talk input mechanism; a second encoder configured for far-talk input captured by a far-talk input mechanism; and an encoder selection layer configured to select at least one of the first and second encoders for use in producing ASR output. The selection is made based on at least one of short-time Fourier transform (STFT), Mel-frequency Cepstral Coefficient (MFCC) and filter bank derived from at least one of the close-talk input and the far-talk input. If signals from both the close-talk input mechanism and the far-talk input mechanism are present for a speech segment, the encoder selection layer dynamically selects between the close-talk encoder and the far-talk encoder to select the encoder that better recognizes the speech segment. An encoder-decoder model is used to produce the ASR output.
Legal claims defining the scope of protection, as filed with the USPTO.
2. The system according to claim 1, wherein the encoder selection layer dynamically switches between the first encoder and the second encoder to select an encoder that better recognizes the speech segment.
3. The system according to claim 1, wherein only one of the quality of the close-talk input or the quality of the far-talk input is used to select the first encoder, the second encoder, or both for producing ASR output.
4. The system according to claim 1, wherein the far-talk input mechanism is a multi-channel input mechanism for capturing a multi-channel far-talk input signal.
6. The system according to claim 1, wherein the first type of input device is different from the second type of input device.
8. The system according to claim 1, wherein the first type of input device comprises a headphone or an MP3 recorder.
9. The system according to claim 8, wherein the second type of input device comprises a microphone array.
10. The system according to claim 9, wherein the encoder selection layer is configured to select the first encoder, the second encoder, or both based on the quality of the close-talk input captured by the headphone or the MP3 recorder, the quality of the far-talk input captured by the microphone array, or both.
12. The computer-implemented method according to claim 11, wherein the encoder selection layer dynamically switches between the first encoder and the second encoder to select an encoder that better recognizes the speech segment.
13. The computer-implemented method according to claim 11, wherein only one of the quality of the close-talk input or the quality of the far-talk input is used to select the first encoder, the second encoder, or both for producing ASR output.
14. The computer-implemented method according to claim 11, wherein the far-talk input mechanism is a multi-channel input mechanism for capturing a multi-channel far-talk input signal.
16. The computer-implemented method according to claim 11, wherein the encoder selection layer is configured to select the first encoder and then switch to the second encoder based on the close-talk input being followed by the far-talk input.
18. The computer-implemented method according to claim 11, wherein the first type of input device comprises a headphone or an MP3 recorder.
19. The computer-implemented method according to claim 18, wherein the second type of input device comprises a microphone array.
20. The computer-implemented method according to claim 19, wherein the encoder selection layer is configured to select the first encoder, the second encoder, or both based on the quality of the close-talk input captured by the headphone or the MP3 recorder, the quality of the far-talk input captured by the microphone array, or both.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
June 22, 2021
May 7, 2024
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.